Free ISACA AAISM Practice Test & Real Exam Questions, Page 4

Question #46

Which of the following AI-specific security monitoring metrics BEST enables the real-time detection of attempts to evade an AI model?

A. Anomalous input leading to unexpected model output

B. Degree of drift in comparison with the initial model baseline

C. Decrease in the model's availability and performance results

D. Decrease in the model's feature importance stability and explainability scores

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #47

Which of the following BEST reduces the risk of exposing sensitive data through the output of large language models (LLMs) in applications?

A. Implementing data sanitization techniques

B. Encrypting data in transit and at rest

C. Conducting adversarial testing

D. Enforcing least privilege access

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #48

The PRIMARY benefit of implementing moderation controls in generative AI applications is that it can:

A. Filter out harmful or inappropriate content

B. Optimize the model's response time

C. Ensure the generated content adheres to privacy regulations

D. Increase the model's ability to generate diverse and creative content

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #49

As organizations increasingly rely on vendors to develop AI systems, which of the following is the MOST effective way to monitor vendors and ensure compliance with ethical and security standards?

A. Allowing vendors to self-attest ethical AI compliance and implement benchmark monitoring

B. Conducting regular audits of vendor processes and adherence to AI development guidelines

C. Requiring vendors to monitor their adherence to ethics and security standards

D. Mandating that vendors share source code and AI documentation with the contracting party

Discussion 0

Correct Answer: B Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #50

When designing self-driving vehicles, conformity assessment is PRIMARILY needed to:

A. protect against unfair decisions.

B. assess compliance against AI regulations and frameworks.

C. test for security flaws before the AI system goes live.

D. verify that the AI system operates reliably.

Discussion 0

Correct Answer: B Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #51

Which of the following is the MOST effective strategy for penetration testers assessing the security of an AI model against membership inference attacks?

A. Disabling AI model logging to reduce noise during testing

B. Measuring AI model accuracy on the test set

C. Generating synthetic data to replace the training data

D. Analyzing AI model confidence scores to indicate training data

Discussion 0

Correct Answer: D Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #52

Which of the following is the BEST way to secure the deployment of an AI agent?

A. Conduct periodic reviews of agentic behaviors.

B. Develop an agent-related incident response process.

C. Implement monitoring of agentic actions.

D. Ensure sufficient levels of access for the agent.

Discussion 0

Correct Answer: C Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #53

An organization uses an AI model to optimize traffic signals and predict congestion. Due to misconfiguration and lack of validation, the system over-prioritizes low-traffic areas, causing gridlock in major cities. Which of the following should be done FIRST to address this issue?

A. Roll back the AI system to a previous version.

B. Continue running the AI model and deploy a secondary system.

C. Retrain the model in real time using new traffic data.

D. Disable the AI system and issue a public statement.

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #54

Which of the following datasets is MOST appropriate to use when tuning an AI model's hyperparameters to improve performance?

A. Synthetic

B. Validation

C. Training

D. Testing

Discussion 0

Correct Answer: B Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #55

Which of the following is BEST to review in order to detect issues related to data used throughout an AI system's life cycle?

A. Data retention policies

B. User access logs

C. Data quality requirements

D. Data flow diagrams

Discussion 0

Correct Answer: D Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #56

An organization is facing a deepfake attack intended to manipulate stock prices. The organization's crisis communication plan has been activated. Which of the following is MOST important to include in the initial response?

A. Conduct employee awareness training on recognizing deepfake videos and audio

B. Conduct a detailed forensic analysis to identify the source of the deepfake

C. Engage with brand monitoring services to track social media activity

D. Provide clarifying information in a pre-approved public statement

Discussion 0

Correct Answer: D Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #57

An organization has implemented a natural language processing model to respond to customer questions when personnel are not available. A pre-implementation security assessment revealed attackers could access sensitive company data through a chat interface injection attack. Which of the following is the BEST way to prevent this attack?

A. Implementing input validation and templates

B. Manually reviewing AI model outputs

C. Conducting regular information security audits

D. Ensuring continuous monitoring and data tagging

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #58

Which AI data management technique involves creating validation and test data?

A. Annotating

B. Training

C. Splitting

D. Learning

Discussion 0

Correct Answer: C Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #59

Which of the following is the MOST critical success factor for an AI implementation project?

A. Obtaining senior management buy-in

B. Developing and using model cards

C. Ensuring AI risk is captured in the risk register

D. Mapping data throughout the life cycle

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Question #60

To process customer feedback for business decisions, an organization purchases an AI-powered sentiment analysis service implemented as a black box API. Which of the following represents the GREATEST security risk to the business?

A. The service may secretly exfiltrate customer feedback data for its own purposes.

B. The API endpoint for the service could be unavailable due to a distributed denial of service (DDoS) attack.

C. The service could be subjected to data poisoning attacks that causes sentiment analysis to be manipulated.

D. The organization's API key for the service could be stolen and used by an unauthorized party.

Discussion 0

Correct Answer: A Vote an answer

Explanation: Only visible for Pass4Leader members. You can sign-up / login (it's free).

Free ISACA AAISM Practice Test & Real Exam Questions

Download Free ISACA AAISM Demo